Parallel Reduction (Sum) Visualizer

Parallel Reduction (Sum)

parallelPRAM-EREW

An algorithm that combines a collection of values using an associative binary operator (like addition or maximum) in parallel, constructing a binary tree of evaluations in O(log N) time steps.

SIMULATION STEP PROGRESSStep 1 / 1

SPEED (50%)

SIZE (8)

CORES (8)

NETWORK TOPOLOGY

💡 Tip: Enter integers between 1 and 99. Max size is 16.

Code Workspace

1procedure ParallelReduction(A: array of size N)

2 for d := 1 to log2(N) do

3 for each processor P_i where i < N / (2^d) in parallel do

4 A[2 * i * 2^(d-1)] := A[2 * i * 2^(d-1)] + A[(2 * i + 1) * 2^(d-1)]

5 end parallel-for

6 synchronize / barrier()

7 end for

8 return A[0]

9end procedure

Currently highlights executing branch

RUNTIME COST METRICS

Comparisons0

Memory Writes0

Memory Reads0

Elements Swapped0

Core Capacity:8 Cores

Total Work Done:5 operations

Critical Path Span:2 levels

Observed Speedup:2.5x

Efficiency Bound:31%

PROCESSOR ALLOCATION EFFICIENCY

THEORETICAL ANALYSIS & RESEARCH

Conceptual Focus: An algorithm that combines a collection of values using an associative binary operator (like addition or maximum) in parallel, constructing a binary tree of evaluations in O(log N) time steps.

Real-world Industrial Applications:

Computing sums, dot products, mins, or maxes of massive vectors in GPUs
MapReduce frameworks where data needs aggregate stats